Calibration drift rate & interval calculator
Enter a drift rate, or two calibrations, plus the tolerance margin you have left and a safety factor. Get an estimate of how long until the reading risks going out of tolerance and a recommended calibration interval. Everything runs in your browser.
Your numbers
How much the reading moves per period, in the same units as your margin. Sign is ignored; magnitude is what counts.
The time unit your drift rate (and elapsed time) is expressed in. Results are shown in months.
How far the reading may move before it is out of tolerance, in the same units as the drift. Often the spec limit minus where it sits now, or the tolerance half-width.
Fraction of the time-to-limit to actually schedule, so you recalibrate before the edge. 0.7–0.8 is a common starting point.
Result
Drift-based intervals, in plain terms
Why drift sets an interval
An instrument that drifts has a built-in clock. If it leaves the last calibration reading true and creeps a little each month, there is a date when it will cross the tolerance limit and start reporting out of specification. Recalibrating before that date keeps every measurement in between trustworthy. The faster the drift and the tighter the margin, the sooner that date arrives, and the shorter the interval has to be.
The projection
This tool draws a straight line. It takes the room the reading has left to move (the available margin) and divides it by the drift rate to get the time to reach the limit: time‑to‑limit = margin ÷ drift rate. Then it applies a safety factor below 1 so you recalibrate before the line touches the edge: recommended interval = time‑to‑limit × safety factor. If you do not have a rate handy, the two-calibration mode derives one from your as-left and as-found values: drift rate = (as-found − as-left) ÷ elapsed time.
Where the line bends
Real drift is rarely a perfect straight line. It can accelerate as a component ages, reset after an adjustment, or hide inside measurement noise so that two points alone mislead you. A two-point rate is a starting estimate, not a verdict. The honest way to confirm an interval is to look at several as-found results over time and see whether the trend holds. When in doubt, schedule sooner: the cost of an early calibration is small next to the cost of a run of out-of-tolerance measurements found after the fact.
NCSLI RP-1 and ISO/IEC 17025:2017 (clause 6.4.7 on calibration status and intervals) expect intervals to be justified and reviewed against evidence, not set once and forgotten. A drift projection is one input to that justification. For reliability-based or history-based methods, use the calibration interval calculator.
Common questions
-
What is instrument drift?
Drift is the slow, systematic change in an instrument’s reading over time, separate from random noise. A reference standard, gauge, or sensor that read true at its last calibration may read slightly high or low months later. If drift continues unchecked, the reading eventually crosses the tolerance limit and the instrument is out of specification. The whole point of a calibration interval is to recalibrate before that happens.
-
How is a drift rate turned into a calibration interval?
First find how much room the reading has to move before it is out of tolerance (the available margin). Divide that margin by the drift rate to get the time to reach the limit. Then multiply by a safety factor below 1 (0.7 is a common starting point) so you recalibrate before the reading actually reaches the edge. The result is a drift-limited interval: interval = (available margin ÷ drift rate) × safety factor.
-
How do I get a drift rate from two calibrations?
Take the as-left value recorded at the previous calibration (where the instrument sat after any adjustment) and the as-found value at this calibration (where it sat before adjustment), then divide the change by the elapsed time between the two. Drift rate = (as-found − as-left) ÷ elapsed time. Two points give you a first estimate; three or more as-found points reveal whether the drift is steady, accelerating, or just noise.
-
What safety factor should I use?
There is no single mandated value. A factor of 0.7 to 0.8 of the projected time-to-limit is a common, defensible starting point: it builds in margin for the fact that drift is rarely perfectly linear and that a single rate estimate carries uncertainty. Tighten the factor (schedule sooner) for critical measurements, accelerating drift, or sparse history; relax it only with a solid as-found record. Your reliability target and risk tolerance govern the final number.
-
What if my instrument shows no measurable drift?
If the drift rate is effectively zero, the interval is not drift-limited: a linear projection would put the time-to-limit at infinity, which is not a usable answer. That is good news, but the interval should then be set from other factors, such as manufacturer guidance, usage and environment, regulatory requirements, or a reliability-based analysis of your as-found history. Use the calibration interval calculator for those methods.
This calculator is a simple linear-drift projection to help you reason about a calibration interval. It does not set your laboratory’s interval policy or replace NCSLI RP-1, ISO/IEC 17025, or your own as-found history. Confirm any interval against several as-found results and your quality system before you adopt it.
Track drift across every calibration
Axiospec keeps every as-found and as-left reading in one tamper-evident ledger, so the drift trend that justifies your interval is always there. See it on real calibration data, no signup.